Hydroxyapatite [HA] is a crystalline mineral of calcium phosphate with a structural formula of Ca10(PO4)6(OH)2. Fluorapatite may be prepared by fluoridating hydroxyapatite, creating a mineral with the structural formula Ca10(PO4)6F2. Protein-reactive sites on both minerals include pairs of positively charged calcium ions (C-sites) and triplets of negatively charged phosphate groups (P-sites). C-sites interact with proteins via HA calcium chelation by protein carboxyl clusters. C-sites interact with phosphorylated solutes such as DNA, endotoxin, phosphoproteins, and lipid enveloped viruses via HA calcium coordination by solute phosphate residues. Calcium chelation and coordination are sometimes referred to as calcium affinity. P-sites interact with proteins via phosphoryl cation exchange with positively charged protein amino acid residues (Gorbunoff, Analytical Biochemistry 136 425 (1984); Kawasaki, J. Chromatography 152 361 (1985)). Hydroxyapatite is most commonly eluted with phosphate gradients. The strong calcium affinity of phosphate suspends calcium chelation and coordination interactions, while its ionic character suspends phosphoryl cation exchange interactions. Some applications elute hydroxyapatite with combinations of phosphate and chloride salts. Chlorides preferentially elute the phosphoryl cation exchange interaction while having relatively little effect on calcium affinity interactions. (Gagnon et al, Bioprocess International, 4(2) 50 (2006)).
Native hydroxyapatite and fluorapatite can be converted to calcium-derivatized forms by exposure to soluble calcium in the absence of phosphate. (Gorbunoff, Anal. Biochem., 136 425 (1984)). This converts P-sites into secondary C-sites, abolishing phosphoryl cation exchange interactions, increasing the number of C-sites, and fundamentally altering the selectivity of the apatite support. Small alkaline proteins typified by lysozyme (13.7-14.7 Kda, pI 10.7) and ribonuclease (14.7 kDa, pI 9.5-9.8) fail to bind to calcium-derivatized apatites, but most other proteins bind so strongly that even 3 M calcium chloride is inadequate to achieve elution (Gorbunoff). Other chloride salts also fail to achieve elution. Calcium-derivatized apatites are restored to their native forms by exposure to phosphate buffer, at which point they may be eluted by methods commonly applied for elution of native apatite supports.
The effects of different salts on the selectivity of a given apatite are unpredictable. For example, in the absence of phosphate, sodium chloride is unable to elute most IgG monoclonal antibodies from native hydroxyapatite, even at concentrations in excess of 4 moles per liter (Gagnon et al, 2006, Bioprocess International, 4(2) 50). This implies extremely strong binding. In exclusively phosphate gradients, IgG is typically one of the latest eluting proteins, usually requiring 100-150 mM phosphate. This also implies strong binding. When eluted with a combination of lower concentrations of both salts, such as 0.25 M sodium chloride and 50 mM phosphate however, IgG is one of the earliest eluting proteins. Other paradoxes reinforce the point: increasing the sodium chloride concentration in the presence of phosphate, which causes IgG to bind less strongly, has the opposite effect on DNA (Gagnon et al, 2005, Bioprocess International, 3(7) 52-55). Additionally, lysozyme elutes at a higher phosphate concentration than BSA in the absence of sodium chloride but fails to bind in the presence of 1 M sodium chloride.
Ammonium sulfate, sodium sulfate, and other sulfate salts are commonly used for precipitation of proteins, or to cause proteins to bind to hydrophobic interaction chromatography media. They can also be used to enhance binding with biological affinity chromatography media such as protein A, and have even been reported to cause proteins to bind to ion exchangers (Gagnon, 1996, Purification Tools for Monoclonal Antibodies, ISBN 0-9653515-9-9; Mevarech et al, 1976, Biochemistry, 15, 2383-2387; Leicht et al, 1981, Anal. Biochem., 114, 186-192; Arakawa et al, 2007, J. Biochem. Biophys. Met., 70, 493-498). Sulfates have occasionally been reported for elution of ion exchangers at low concentrations for research applications but are seldom exploited in preparative applications due to concerns over protein precipitation (Kopaciewicz et al, 1983, J. Chromatogr., 266 3-21; Gooding et al, 1984, J. Chromatogr., 296, 321-328; Rounds et al, 1984, J. Chromatogr., 283 37-45). None of these methods is an appropriate model for apatites because none of them exploits calcium affinity for binding.
Several authors have concluded that, “The presence of . . . (NH4)2SO4 seems not to affect the elution [of hydroxyapatite].” (Karlsson et al, 1989, in Protein Purification: Principles, High Resolution Methods, and Applications, Chapter 4, ISBN 0-89573-122-3). Even this reference mentions the application of sulfate strictly in the context of phosphate gradients. In the rare cases where alternatives to phosphate as a primary eluting salt have been discussed in the literature, suggestions have included calcium chloride, citrate and fluoride salts, but without mention of sulfates (Gagnon, 1996; Karlsson et al, 1989; Gorbunoff). Other publications indicate that sulfate salts in particular should be unsuitable as primary eluting agents for hydroxyapatite because “ . . . SO3H do[es] not form complexes with calcium.” (Gorbunoff).
Borate salts have been likewise overlooked. Borate is occasionally used in the field of chromatography as a buffering agent at pH values from about 8.8 to 9.8 (pK˜9.24). It is also used infrequently at alkaline pH to modify the charge characteristics of cis-diol compounds to selectively enhance their retention on anion exchangers. In contrast to phosphates, chlorides, and sulfates, all of which exhibit molar conductivities of about 90 mS/cm, a 1 M solution of borate at pH 7 has a molar conductivity of about 9 mS.
Acetates have been compared to chlorides for hydroxyapatite separation of IgG from aggregates and were found to support inferior fractionation (Gagnon et al, Practical issues in the industrial use of hydroxyapatite for purification of monoclonal antibodies, Poster, 22nd national meeting of the American Chemical Society, San Francisco, Sep. 10-14, 2006 <http://www.validated.com/revalbio/pdffiles/ACS_CHT_02.pdf>. Monocarboxylic acid salts have been neglected, and the elution potential of monocarboxylic zwitterions totally so.
Hydroxyapatite is used for purification of antibodies and antibody fragments (Bowles et al, Int. J. Pharmacol., 10 537 (1988); Guerrier et al, J. Chromatography B, 755 37 (2001); Gagnon et al, BioProcess Int., 4(2) 50 (2006)). The column is usually equilibrated and the sample applied in a buffer that contains a low concentration of phosphate. Adsorbed antibodies are usually eluted in an increasing gradient of phosphate salts. Alternatively, they may be eluted in an increasing gradient of chloride salts but both elution formats impose disadvantages on purification procedures. The high phosphate concentration in which antibodies elute in phosphate gradients has strong buffer capacity that may interfere with subsequent purification steps. The high conductivity at which antibodies elute in chloride gradients may also interfere with downstream steps. Both situations require either that the eluted antibody be diluted extensively, or that it be buffer-exchanged, for example by diafiltration, in order to modify the conditions to render the antibody preparation suitable for application to a subsequent purification step. Dilution and buffer exchange have a negative impact on process economics. As a result, apatite chromatography steps are often placed at the end of a purification process. This tends to eliminate them from consideration as capture steps. It also discourages the use of HA as an intermediate step. A further disadvantage of chloride gradients is that the application of chloride to hydroxyapatite causes an uncontrolled reduction of pH. Acidic pH causes destruction of hydroxyapatite and risks adverse affects to antibodies bound to it.
Another limitation of hydroxyapatite with antibody purification is that IgG binding capacity is reduced at elevated conductivity values. This strongly reduces its versatility since the salt concentration of cell culture supernatants and antibody-containing fractions from purification methods such as ion exchange and hydrophobic interaction chromatography, confers sufficient conductivity to reduce the IgG binding capacity of hydroxyapatite to such an extent that it may not be useful for a particular application. This disadvantage can be overcome by diafiltration or dilution of the sample prior to its application to the hydroxyapatite column, but as noted above, these operations increase the expense of the overall purification process. Alternatively, the disadvantage can be ameliorated by using a larger volume of hydroxyapatite, but this increases process expense by requiring larger columns and larger buffer volumes. It also causes the antibody to elute in a larger volume of buffer, which increases overall process time in the subsequent purification step.